AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Beyond Sequential Distance: Inter-Modal Distance Invariant Position Encoding

Lin Chen, Bolin Ni, Qi Yang, Zili Wang, Kun Ding, Ying Wang, Houwen Peng, SHIMING XIANG

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Training-free token reduction for MLLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add DIPE (anchored inter-modal position encoding)MRoPE, Qwen2.5-3B · Long-context VQA (8K distractors), all 19 benchmarks avgaverage accuracy: 44.69 -> 48.79Improved
add DIPEVanilla RoPE, Qwen2.5-3B · Long-context VQA (8K distractors), all 19 benchmarks avgaverage accuracy: 46.31 -> 48.31Improved
add DIPEMRoPE-I, Qwen2.5-3B · Long-context VQA (8K distractors), all 19 benchmarks avgaverage accuracy: 46.50 -> 48.51Improved
add DIPEMRoPE, Qwen2.5-3B · POPE (long-context 8K)accuracy: 74.50 -> 85.58Improved
add DIPEMRoPE-I, Qwen2.5-3B · V* (long-context 8K)accuracy: 50.79 -> 47.64Hurt
add DIPEMRoPE, Qwen2.5-3B · Short-context VQA, all 19 benchmarks avgaverage accuracy: 51.15 -> 51.38No clear effect
add DIPEMRoPE-I, Qwen2.5-3B · Short-context VQA, all 19 benchmarks avgaverage accuracy: 51.49 -> 51.41No clear effect
add DIPEMRoPE, Qwen3-1.7B · Long-context VQA (8K distractors), all benchmarks avgaverage accuracy: 40.21 -> 44.46Improved
add DIPEMRoPE, Qwen2.5-0.5B · Long-context VQA (8K distractors), all benchmarks avgaverage accuracy: 29.51 -> 38.32Improved
add DIPEMRoPE, Qwen2.5-3B · Interleaved image-text test (1K samples)accuracy: 60.1 -> 60.4No clear effect
add DIPEMRoPE, Qwen2.5-3B · Latency, attention block forward at 32Klatency (ms): 13.56 -> 14.31Hurt
add DIPEMRoPE, Qwen2.5-3B · Latency, attention block backward at 1Klatency (ms): 0.83 -> 1.52Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy